Entraides et échanges autour de la technologie Scol - Informations and exchanges on the Scol technology
Vous pouvez changer la langue de l'interface une fois inscrit - You can change the language once registered
You are not logged in.
Copying dll directly is a >very< bad habit, because they can have dependencies with other dll (which in turns could have dependencies with more dlls, etc.).
Your problem is certainly that directX is not installed on the pc you tested on. You can download the last version on misocroft's website.
Otherwise, the other solution is to export using OpenGL, but your materials/shaders have to be tested before, to be sure that they are supported by OpenGL.
Verify that the last driver for the graphic card is installed too.
Congratulations
Scene collision is not supported by OpenSpace3D (look at the dropbox listing all the collisions type that are possible!), because a tree collision works better for most of the meshs, for the same functionalities. As a side node, height-field collision is not supported neither, at least at this time.
So, as far as I understand what you want to do, use a tree collision for the static part of your scene (the road, and other static objects?), and a convex collision (named "shape" in openspace3d) for the car (although a box collision may be sufficient).
Openspace3D uses Newton library for physics, so you can found information about which collision shape is the best for a given usage here:
Newton collision shapes
Argg, arkeon was <again> faster than me , but let's say that's because I'm talking too much
What you need is to add another "partition" (in fact it's a window directory) to scol, and work on it. There's a procedure on the forum, but it's in French, so i'll translate it here for you
You need to modify the "usm.ini" file, which default location is "C:\Program Files\Scol Voyager". You should find those lines in it (they are the default values for partitions):
cache ./Cache 256
disku ./Partition_LocalUsr 0
diska ./Partition_LockedApp 0
logsp ./Logs 0
The number at the end of every partition line is the maximal amount of data that you can store (in Mo i think) on a partition, 0 meaning that there's no restriction of size.
As you can see, every directory is "relative" to a default location: "cache" and "logsp" are relative to "C:\Users\UserName\AppData\Local\Scol Voyager", "disku" is relative to user document directory (default: "C:\Users\UserName\Documents\Scol Voyager"), and "diska" is relative to Scol installation path (default: "C:\Program Files\Scol Voyager").
So, as you can see, a problem could append when changing the user partition "disku" to another drive, as apparently, it uses a relative path to a default location that is under "C:\". Anyway, you can try to change "disku" with a full path ("d:\toto\tutu") to see if it works, but i did not test this manipulation.
Another solution could be to create a "symbolic link" (if you are under win7), using the command "mklink" (with "/D" option for a directory) on the command line.
There's also those translations that ModulaMix had done recently, on the redmine page:
http://trac.arkeon.be/projects/scol/wiki/ScolLanguage
and more specifically, an introduction to OpenSpace3D's plugIt coding:
http://trac.arkeon.be/projects/scol/wik … uginsHowTo
You could also find info at this last page (although there's not much except some code templates):
http://trac.arkeon.be/projects/scol/wiki/Tutorials <edit>Did not saw, but it's the same page as the one mentionned previously by Iri</edit>
The actual documentation sources files are in French, we actually use Google's automatic translation to provide an English version of the documentation...
You can of course publish your translation of the doc on another website, albeit it should be better for us to have a link on our website too.
So, I think it should be better for everybody that you translation on our subversion server, and let us publish it (it will be simpler for maintenance), letting you make a copy on a third server if you wish (as the doc web pages source code is freely available for everybody on the subversion server).
@hebdemnobad In fact, we started to completely change the physic engine (there's a branch on the svn), but we do not have any people to work on it at this time...
@gtavcman For your application, did you saw the "FPS navigation plugit", that could be a good starting point... To try it, you can for example open the "Scene3d" demo, and replace the navigation system with it.
It works on my Seven SP1 / GeForce 8800 GTX pc.
Cool, but two point concerning your modeling:
I assume that you mirrored an half of the model, cause there's a little overlap at the center of it...
And you may have to add some vertex and remap your uv mapping for the torso part, to limit the stretching of the texture.
Cool, nice pic
Have a smile next time
The explication is that our physic engine is not deterministic (like 99% of real time other physic engines), which means that playing the same simulation 2 times will not render the same result. It's the case in <put your favorite multiplayer game here> too, cause for example we don't care if the bottle I just come shooting on moves exactly the same way as it will on the distant computer my friend plays with.
Anyway, what you are asking for is possible with openspace3d, but you'll need two different scenes, a "master scene" with physics on, and some "clients scene" with physic off, just updating object positions.
No, he mean that he want to use a periodic table image as marker...
You can't do that with the AR library OpenSpace uses... The ones that let you use any image as marker are not free at all...
Bon ben on est partit dans le troll de toutes façon, ça fait un bail qu'il s'agit plus de syspack
Et encore, je t'évite la liste des trucs que le compilo autorise alors qu'il ne devrait pas... Allez, juste deux de mémoire pour le plaisir:
- Avec Bilou, tu as le droit de déclarer une fonction membre d'une classe en précisant dans quelle classe tu est:
class MaClasse
{
public:
void MaClasse::UneFonction();
// ********* <-- c'est pas dans la norme, incorrect, mais ça passe en msvc
};
- Avec Bilou, le compilo il te fait croire qu'il est capable de deviner de quelle implémentation d'un template tu parles quand tu utilises un de ses membres:
template <typename T> class MaClasse
{
private:
int var;
public:
void UneFonction();
};
template <class T> void MaClasse<T>::UneFonction()
{
var = 10;
// *** <-- incorrect, ça compile avec msvc, mais tu as une chance que ça pète à l'exe, car on sait pas si c'est MaClasse<int>::var ou MaClasse<char>::var (ou un autre paramètre template, on s'en fout) de qui il s'agit ici.
MaClasse<T>::var = 10;
// La norme... le pire, c'est que ça compile sous msvc... mais c'est pas obligatoire!!!
}
Promis, après j'arrête . Y'a quand même des trucs de bien... Le debugger est inégalé par exemple.
Ben oui, mais c'est le cas... me parle pas de norme, j'ai regardé le C++11 (pas C mais bon), validé en aout, plein de nouveaux trucs dans le langage. Résultat: maj gcc déjà presque complète, béta de msvc11 -> environ 3% des nouveautés qui seront intégrées (j'ai trouvé ça sur un blog d'un des dev de microsoft).
Donc pour pas de warning, faut pas utiliser la fonction de la norme C... <edit>sous windoze</edit>
Toutes les fonction "_s" de microsoft, les "secures" comme ils appellent ça, prennent toutes un size_t supplémentaire en param qui correspond à la taille maxi du "char*" que tu passes.
Sauf que effectivement, si on ajoute ce param à un sprintf (non "secure" selon Bilou) pour faire un sprintf_s, on à la même chose qu'un snprintf, (qui lui est "secure" selon Redmond... Heu attendez, zètes sûr, y'a un truc qui m'échappe ...)!
De toutes manières, si tu calcules mal la taille, c'est pas plus secure, et comme les erreurs d'overflow avec ces fonctions sont quasiment toujours du à un mauvais calcul de cette taille ("j'ai déjà ajouté +1 pour le \0 ou pas?"...), c'est de la br******e microsoftienne.
Le truc, c'est que sous Windaube, les fonctions sans "_s" sont dépréciées, et il était sujet de les virer à un moment donné (heureusement, ce n'est plus d'actualité, il semblerait qu'elle seront conservées, mais ça générèe plein de warning dégueus quand tu les utilises). Donc, dans ce cas, ce serait de faire un define, SCOL_SNPRINTF par exemple, qui prenne soit le snprintf normal pour gcc, ou l'autre sprintf_s pour msvc pour éviter les warnings.
Voila. Vive Microsoft, Vive le royaume de Redmond, et longue vie à Bilou (si avec ça, ça part pas en troll ).
Oui, un petit define pour une fonction qui n'existe pas sur tout les plateformes (je pense aux fonctions dites "sécurisée" de la lib std de msvc++, genre printf_s, qui en fait est pas du tout plus sécurisé si le développeur fait nawak) pour chaque compilo peut être utile dans certains cas.
Adding the following line
plugin plugins/libsqlite3.dll
without "ScolLoadPlugin ScolUnloadPlugin" shoud be sufficient on windows platforms. Making uniform the dll entry/exit point for every plugin is a recent change of the scol.exe launcher.
There is at least two techniques to render 3D scene in a navigator without using WebGL:
- Use a proprietary plugin, that will inform the user he needs to install some new components in his computer.
- Use flash or java (which are plugins too, but installed on most of computers) as a "wrapper" that will download your plugin (your 3D engine if you prefer) without asking anything.
The second seems better, as the user does not do anything... but, it installs software on the user pc without warning him (except the navigator security message, which do not say explicitly "hey, I'm going to install a software on your pc and you don't know what it is"). The security concerns should always be taking in count in web technologies (even if it makes the life of the users <a little> more complicated).
So, I think that we will stay on the first solution...
Sorry, but I do not have tutorial links at all...
But, for material files examples, you can check Ogre's wiki page (for example, http://www.ogre3d.org/tikiwiki/Materials).
The mandatory script manual, for understanding material files keywords: http://www.ogre3d.org/docs/manual/manual_14.html#SEC23 (and the following pages).
You can also use a material editor inside the modeler, using OgreMax or Blender2Ogre (integrated material editor is a OpenSpace3D's planned functionnality, but it does not exists at this time...), or search in ogre's forum (http://www.ogre3d.org/) for a third party material editor (I know this one, but I never use it, so I do not know how reliable it is).
Finally, a search on ogre's forum for the effect you need is often successfull.
Still not clear for me . I'd like to help, but if i do not see exactly what you wish to do, how could I
?
Do you whant to change the loading object when the user "clic" on the right arrow like on the video? Please try to explain precisely what you want !
And, you did not explain in which way you didn't succeed to load a 3D object... We are not magicians, and further less telepaths, so please, try to reformulate your question(s?) ...
when kinect finished tracked my pose
What do you mean exactly?
i cant load my object on the screen
Not sure about that neither, but did you mean that you have problems to add your mesh to the scene?
Je n'avais pas vu ta seconde question, la réponse est mitigée:
Pour le positionnement des objets, ce serait en effet faisable, mais il faudrait un plugin Ogre de chargeur de fichier ".xos", ce qui n'est pas bien compliqué à faire, je pense qu'il y en a pour 1 journée ou deux en se basant sur le chargeur de fichier ".scene" existant sur le wiki de Ogre.
Par contre, et bien évidemment, pas d'interactions exportées dans Ogre, je pense qu'il n'est pas la peine de préciser pourquoi